Human kallikrein-related peptidase 6 (KLK6) was cloned as a putative class II tumor suppressor based on its inactivated expression in metastatic breast cancer. Here, we investigated the mechanism(s) underlying the silencing of KLK6 gene in metastatic breast cancer and its putative implications for tumor progression. We present evidence that tumor-specific loss of KLK6 expression is due to hypermethylation of specific CpGs located in the KLK6 proximal promoter. Methylation-dependent binding of methyl CpG-binding protein 2 and the formation of repressive chromatin mediated by localized histone deacetylation are critical components of KLK6 silencing in breast tumors. Re-expression of KLK6 in nonexpressing MDA-MB-231 breast tumor cells by stable cDNA transfection resulted in marked reversal of their malignant phenotype, manifested by lower proliferation rates and saturation density, marked inhibition of anchorage-independent growth, reduced cell motility, and their dramatically reduced ability to form tumors when implanted in severe combined immunodeficiency mice. Interestingly, inhibition of tumor growth was observed at physiologic concentrations of KLK6, but not when KLK6 was highly overexpressed, as observed in a subset of breast tumors. Differential proteomic profiling revealed that KLK6 re-expression results in significant down-regulation of vimentin which represents an established marker of epithelial-to-mesenchymal transition of tumor cells and in concomitant up-regulation of calreticulin and epithelial markers cytokeratin 8 and 19, indicating that KLK6 may play a protective role against tumor progression that is likely mediated by inhibition of epithelial-to-mesenchymal transition. We suggest that KLK6 is an epigenetically regulated tumor suppressor in human breast cancer and provide ways of pharmacologic modulation. [Cancer Res 2009;69(9):3779–87]

The cDNA encoding kallikrein-related peptidase 6 (KLK6) was originally identified by mRNA differential display as being highly overexpressed in a primary breast tumor, but completely inactivated in its lung metastasis and the vast majority of metastatic breast cancers (1). Based on this expression pattern, it was suggested that the encoded protein—a novel serine protease—may function to protect against tumor progression and that it is likely deregulated at the transcription level; therefore, it represents a putative class II tumor suppressor (1). However, the putative function(s) of KLK6 in cancer are yet to be elucidated. Remarkably, aberrant expression of KLK6 gene has been implicated in Alzheimer's and Parkinson's disease (2). In addition, KLK6 is involved in enhanced proteolysis of myelin basic protein associated with multiple sclerosis and central nervous system inflammation (35); therefore, it represents a potential therapeutic target for pharmacologic intervention.

At the protein level, the enzymatic activity of KLK6 is controlled by a self-regulatory mechanism (6) that is further supported by the crystal structures of the KLK6 zymogen (7) and mature enzyme (3). The transcription and tissue-specific expression of KLK6 are likely regulated by multiple mechanism(s) that include alternative splicing of coding exons and utilization of alternative intronic promoters (P2 and P3), in addition to the 5′ upstream promoter (P1) that encodes the classic transcript (8, 9).

Here, we investigated the mechanism(s) underlying the aberrant expression of KLK6 in breast cancer cells and its putative implications for tumor progression. Our results provide direct molecular evidence for the functional significance of DNA methylation and chromatin structure with respect to the inactivation of KLK6 expression in metastatic breast cancer. Epigenetic silencing of KLK6 suggested a tumor suppressor role for the encoded protein. Consistently, we show for the first time that re-expression of KLK6 in the highly tumorigenic MDA-MB-231 breast tumor cell line at concentrations similar to those observed in normal mammary epithelial cells and tissues results in slower proliferation rates, significantly decreased in vitro tumorigenicity, and in vivo orthotopic tumor formation in severe combined immunodeficiency (SCID) mice. Differential proteomic profiling revealed that KLK6 reactivation results in remarkable down-regulation of vimentin with concomitant up-regulation of epithelial markers and other proteins associated with tumor reversion. Taken together, our data show for the first time that KLK6 exerts a protective role against breast cancer progression that is likely mediated by inhibition of epithelial-to-mesenchymal transition.

Materials. Synthetic oligonucleotides were purchased from MWG-BIOTECH, 5-aza-2′-deoxycytidine (5-aza-dC) and trichostatin A (TSA) were from Sigma, [α-32P]dCTP from Amersham, and antibodies were from Santa Cruz Biotechnology. Anti–methyl CpG-binding protein 2 (MeCP2) antibody was from Abcam and anti–acetylated histone 4 was from Upstate.

Cell culture and treatments. Cell lines were obtained from American Type Culture Collection and maintained in RPMI supplemented with 10% fetal bovine serum (Life Technologies). Human mammary epithelial cells (HMEC) were obtained from Clonetics and maintained in MEGM (Clonetics). For treatments, cells were seeded in 100 mm dishes and grown for 24 h to reach 20% to 30% confluence. Then, 5-aza-dC was added to final concentrations of 50 or 100 μmol/L and cells were incubated for 48 h followed by another 48 h in fresh medium and finally collected for extraction of total cellular RNA using RNeasy (Qiagen). TSA was added to final concentrations of 50, 100, or 200 nmol/L.

Cytotoxicity assay. Cells (7 × 104) were plated on 24-well plates with 1 mL medium and allowed to adhere for 24 h in order to reach 20% to 30% confluence. 5-aza-dC was added at various concentrations and cells were incubated for another 72 h. Then, 100 μL of 5 mg/mL of 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) solution was added and cells were further incubated for 2 h. Medium was removed, the insoluble formazan crystals were dissolved in 1 mL of 100% DMSO, and absorbance was measured at 570 nm. Background controls included medium incubated with MTT.

RNA extraction and reverse transcription-PCR. RNA was extracted with RNeasy and treated with DNase I (Qiagen). Reverse transcription was carried out with Sensiscript (Qiagen) using 2 μg of total RNA and cDNAs were PCR-amplified with gene-specific primers: KLK6S (5′-GGAGGAATTCCAGCAGGAGCGGCCATG-3′) and KLK6AS (5′-TGTCTCGAGTCAGGGTCACTTGGCCTG-3′) for KLK6, VIMS (5′-GGCTCAGATTCAGGAACAGC-3′) and VIMAS (5′-CTGAATCTCATCCTGCAGGC-3′) for vimentin, ECADS (5′-TGAAGGTGACAGAGCCTCTGGAT-3′) and ECADAS (5′-TGGGTGAATTCGGGCTTGTT-3′) for E-cadherin, and ACTINS (5′-ACAATGAGCTGCGTGTGGCT-3′) and ACTINAS (5′-TCTCCTTAATGTCACGCACGA-3′) for β-actin.

Genomic bisulfite sequencing. Genomic DNA was extracted with GeneElute (Sigma). To convert unmethylated cytosines to uracils, 1 μg of genomic DNA was treated with bisulfite using CpGenome (Intergen) and recovered in 30 μL of TE buffer. Bisulfite-treated DNAs were amplified by PCR with methylation-independent primers: 5′-GTTAGAGATTGTAAAGGAGGATTGT-3′ and 5′-CACACCTACCCATAAATCCCTCTAT-3′. PCR products were cloned into pCR4-TOPO (Invitrogen). Multiple clones were isolated and sequenced.

Chromatin immunoprecipitation. Cells were grown to 80% to 100% confluence. Chromatin was crosslinked with 1% formaldehyde for varying time intervals as described (10). DNA was purified and dissolved in 100 μL of TE. The KLK6 (−259, +120) sequence was amplified by hot PCR using 5 μL of DNA template purified after chromatin immunoprecipitation (ChIP), 30 pmol of KLK6-specific primers: 5′-ATGGAAGATCTCTCCCCTCCATGGCCAGG-3′ (forward) and 5′-GAGACAGCTACAGCGTGTGTCACC-3′ (reverse), 200 μmol/L of deoxynucleotide triphosphates, 0.1 μL of [α-32P]dCTP, and 1 unit per reaction of Taq Polymerase (Promega). PCR products were resolved on 6% acrylamide gels and visualized with Phosphor Imager (Molecular Dynamics SSD).

Western blot. Cell lysates (40 μg) were resolved by 12% SDS-PAGE and blotted on a polyvinylidene difluoride membrane (Millipore) that was blocked and incubated with appropriate antibodies (1:1,000). Then, it was washed in PBS-T and incubated with horseradish peroxidase–labeled secondary antibody (1:4,000). Specific immunoreactive bands were detected with enhanced chemiluminescence (Pierce).

Expression constructs. The cDNA encoding preproKLK6 was PCR-amplified from full-length KLK6 cDNA using primers KLK6S and KLK6AS and cloned into pcDNA3.1(+) (Invitrogen). Plasmid DNAs were purified and confirmed by DNA sequencing.

Stable transfections. MDA-MB-231 cells were grown in 100 mm dishes for 24 h and transfected with Polyfect (Qiagen). Polyfect was removed after 48 h and fresh medium containing 0.5 mg/mL of G418 was added for selection. Individual colonies of stably-transfected cells were picked after 3 weeks. The concentration of KLK6 in serum-free conditioned media was measured by ELISA (11).

Growth curves and soft agar assay. For growth curves, 2.7 × 104 cells were seeded into six-well plates and counted on days 1, 3, 5, 7, and 9. Determination of population doubling time was performed by plotting ln(cell number) versus time (days) for the logarithmic phase of growth and fitting the data according to the least squares mean method. A soft agar assay was performed as previously described (12). Colonies were stained with 1 mg/mL of MTT for 24 h and counted in double-blinded experiments. Results and corresponding SDs were derived from three independent experiments.

Wound scratch assay. Cells (4 × 105) were seeded in six-well plates and left to grow to confluence. The monolayer was wounded by scraping a line across the well with a sterile blue pipette tip. Cells were washed with PBS and refreshed with RPMI supplemented with 10% fetal bovine serum. After 24 h, plates were photographed at the marked spots.

Proteomic profiling. Cell lysates were prepared from 108 cells and resolved by two-dimensional PAGE. Differentially expressed proteins were identified by mass spectrometry. Lysates were analyzed three separate times. Results were confirmed in two independent experiments.

In vivo tumor formation. Cells (2 × 106) were resuspended in 200 μL of PBS and injected bilaterally into the mammary fat-pad of 6-week-old female SCID mice. Mice were examined on alternate days for the presence of palpable tumors. Tumors were allowed to grow for the indicated times and tumor sizes were measured double-blinded. Then, mice were sacrificed and photographed. Expression of KLK6 in dissected tumor specimens was verified by reverse transcription-PCR (RT-PCR) and Western blot (data not shown). Tumor volumes were calculated using the formula: 1/2 × height × width × length. Experiments were conducted according to the guidelines of our institutions for animal handling.

Aberrant expression of KLK6 gene in breast cancer. Differential expression of KLK6 in normal and tumor mammary cell lines was confirmed by semiquantitative RT-PCR. Protein concentrations were measured in cell culture supernatants by a sensitive and specific ELISA (11). For investigation of the molecular mechanism(s) underlying the aberrant KLK6 expression patterns observed in breast cancer, a panel of cell lines was selected which included HMECs normal human mammary epithelial cell strain (12 μg/L KLK6 protein, +KLK6 mRNA), T47D (0.06 μg/L KLK6, +/−KLK6 mRNA) and MDA-MB-231 (0.00 μg/L KLK6, −KLK6 mRNA), two nonexpressing tumor cell lines, and MDA-MB-468, a breast tumor cell line that highly overexpresses KLK6 (400 μg/L KLK6, +++KLK6 mRNA).

Reactivation of KLK6 expression. Treatment of MDA-MB-231 and T47D breast tumor cell lines with 5-aza-dC, an inhibitor of DNA methyltransferases, resulted in significant induction of KLK6 expression (Fig. 1A), implicating genomic DNA methylation in KLK6 silencing. As depicted in Fig. 1B, TSA that acts as a potent inhibitor of histone deacetylases (HDAC) up-regulated KLK6 expression dose-dependently in MDA-MB-231 although to a lesser extent than 5-aza-dC. No change in KLK6 expression was observed upon treatment of T47D cells with TSA. These findings indicated that epigenetic mechanisms likely underlie the inactivation of KLK6 in breast cancer cells and pointed to the predominant role of DNA methylation over histone deacetylation in KLK6 silencing. Reactivation of KLK6 expression was also detected in various breast cancer cell lines using lower concentrations of 5-aza-dC (3–10 μmol/L) for a longer period of time (6 days; data not shown). No signs of severe cytotoxicity were observed in the presence of 100 μmol/L of 5-aza-dC, indicating that reactivation of KLK6 expression is most likely associated with DNA demethylation, although it cannot be excluded that the observed up-regulation might also be due to a secondary effect involving cytotoxic mechanisms. As depicted in Fig. 1C, even after 3 days of treatment with 200 μmol/L of 5-aza-dC, ∼65% of MDA-MB-231 cells were viable. It should be noted that similar 5-aza-dC concentrations were used previously for the induction of genomic DNA hypomethylation in breast cancer cell lines (13).

Figure 1.

KLK6 expression is induced in breast cancer cell lines by DNA methyltransferase and HDAC inhibitors. Identification of methylated CpG dinucleotides in the 5′ upstream sequence of KLK6. A, MDA-MB-231 and T47D cells treated with 5-aza-dC. B, MDA-MB-231 cells treated with TSA. Gene expression was analyzed by RT-PCR. C, no signs of severe cytotoxicity were observed after incubating MDA-MB-231 cells with 5-aza-dC for 72 h. D, methylation of CpGs was analyzed by genomic bisulfite sequencing of HMECs, MDA-MB-468, T47D, and MDA-MB-231 cells. Schematic representation of CpGs (thin vertical lines). Each row represents one sequenced allele. Arrows, transcription start sites of P1 (+1) and P2 (+907) alternative KLK6 promoters. Positions of CpGs analyzed here (lollipops) are given on the horizontal open bar. Ovals, putative Sp1 binding sites identified by SignalScan (http://thr.cit.nih.gov/molbio/signal). Each circle represents a CpG dinucleotide: methylated (•) or unmethylated (○). Numbering was based on GenBank sequence AY804248 of KLK6 P1 promoter.

Figure 1.

KLK6 expression is induced in breast cancer cell lines by DNA methyltransferase and HDAC inhibitors. Identification of methylated CpG dinucleotides in the 5′ upstream sequence of KLK6. A, MDA-MB-231 and T47D cells treated with 5-aza-dC. B, MDA-MB-231 cells treated with TSA. Gene expression was analyzed by RT-PCR. C, no signs of severe cytotoxicity were observed after incubating MDA-MB-231 cells with 5-aza-dC for 72 h. D, methylation of CpGs was analyzed by genomic bisulfite sequencing of HMECs, MDA-MB-468, T47D, and MDA-MB-231 cells. Schematic representation of CpGs (thin vertical lines). Each row represents one sequenced allele. Arrows, transcription start sites of P1 (+1) and P2 (+907) alternative KLK6 promoters. Positions of CpGs analyzed here (lollipops) are given on the horizontal open bar. Ovals, putative Sp1 binding sites identified by SignalScan (http://thr.cit.nih.gov/molbio/signal). Each circle represents a CpG dinucleotide: methylated (•) or unmethylated (○). Numbering was based on GenBank sequence AY804248 of KLK6 P1 promoter.

Close modal

Identification of methylated CpG dinucleotides. Biocomputational analysis using CpGplot (European Bioinformatics Institute) identified no typical CpG islands in 12 kb of KLK6 genomic sequence (GenBank: AF149289). Interestingly though, recent data suggested that gene silencing by DNA methylation does not necessarily require a CpG island (14). Therefore, we searched for methylated cytosines at CpG sites, in association with inactivated KLK6 expression. The 5′ upstream KLK6 (−105, +164) sequence schematically shown in Fig. 1D, which flanks the transcriptional start site of the P1 promoter (15), was selected for bisulfite sequencing analysis based on its high C+G content of 68% and high density of CpG dinucleotides, i.e., 15 CpGs in a region of 269 bp, and the observed/expected ratio of 0.48. Using methylation-independent primers, PCR amplification was carried out to generate a 319 bp amplicon that contained the above CpG dinucleotides. For each cell line, multiple PCR amplicons were cloned and sequenced. Based on the obtained sequences, shown in Fig. 1D, methylated cytosines occurred at positions −72, −64, −56, −53, −35, +3, +14, +97, +114, and +127. Interestingly, the identified CpGs were completely unmethylated in the MDA-MB-468 cell line that overexpresses KLK6. On the other hand, in MDA-MB-231 and T47D cell lines that are characterized by complete loss of KLK6 expression, the identified cytosines were fully methylated. Analysis of a large number of mammary tumor cell lines revealed a direct correlation of KLK6 mRNA levels with the extent of methylation of the identified CpGs (data not shown; Supplementary Fig. S1). As shown in Fig. 1D, HMECs that express relatively low KLK6 mRNA and protein levels were methylated to much lower levels and contained a heterogeneous mixture of methylated and unmethylated alleles. This result indicates that tissue-specific expression of KLK6 may in part be determined by the extent of genomic DNA methylation, as has been well-documented for maspin (16). The methylation status of the identified CpGs was analyzed in tissue specimens obtained from patients diagnosed with breast cancer and normal controls. Correspondingly, similar methylation patterns of the identified CpGs as those described above for tumor cell lines and normal mammary cell strains were detected in clinical tissue specimens (Supplementary Fig. S1).

Methylated CpGs are localized in the KLK6 proximal promoter. Using ChIP assay, we showed that components of the basal transcription machinery, i.e., TATA-binding protein and RNA PolII, are bound to the KLK6 sequence from −259 to +120 in the MDA-MB-468 KLK6-expressing cells but not in the MDA-MB-231 nonexpressing cells (Fig. 2A). For optimization of crosslinking conditions, increasing chromatin-DNA crosslinking times were tested (top). Optimal crosslinking was achieved in 45 min, whereas shorter incubations (15 and 30 min) led to DNA-protein crosslinks that were not stable during subsequent incubation with proteinase K and yielded no detectable PCR products. Prolongation of incubation with formaldehyde may result in changes in the tertiary structure of proteins with implications for antibody recognition. It was found that treatment of MDA-MB-231 cells with TSA causes recruitment of the basal transcription machinery to the KLK6 promoter probably due to acetylation of core histones (Fig. 2A) and activates transcription and KLK6 expression as shown in Fig. 1B, thus, implicating suppressive chromatin structure in the silencing of KLK6 in breast cancer cells.

Figure 2.

Repression of KLK6 expression in breast cancer cells is mediated by methylation-dependent binding of MeCP2. Recruitment events associated with constitutive and inducible KLK6 transcription in MDA-MB-468 and MDA-MB-231 breast cancer cells, respectively, were analyzed by ChIP using antibodies against histone H4 acetylated tail, MeCP2, and histone acetyltransferases CBP and GCN5. A, histone deacetylation upon treatment of MDA-MB-231 cells with 200 nmol/L of TSA reactivated KLK6 transcription by recruiting proteins from the basal transcription machinery on KLK6 chromatin. Crosslinked chromatin prepared from MDA-MB-468 and MDA-MB-231 cells was immunoprecipitated with specific antibodies (right). Increasing crosslinking times (top). Immunoprecipitates were subjected to PCR using a primer pair spanning the region containing the methylated CpGs (Fig. 1D). Aliquots of chromatin taken before immunoprecipitation were used as “Input” controls. B, recruitment of MeCP2, histone acetyltransferases, and acetylation status of H4 at the KLK6 CpG-containing chromatin was shown by ChIP. In vivo binding of Ac-H4, MeCP2, CBP, and GCN5 was examined in KLK6-positive MDA-MB-468 and in KLK6-negative MDA-MB-231 cells (left). KLK6 chromatin is enriched in Ac-H4 upon HDAC inhibition with TSA (right). C, release of the MeCP2 repressor and marked increase of Ac-H4 binding to KLK6 chromatin following treatment of T47D cells with 100 μmol/L of 5-aza-dC.

Figure 2.

Repression of KLK6 expression in breast cancer cells is mediated by methylation-dependent binding of MeCP2. Recruitment events associated with constitutive and inducible KLK6 transcription in MDA-MB-468 and MDA-MB-231 breast cancer cells, respectively, were analyzed by ChIP using antibodies against histone H4 acetylated tail, MeCP2, and histone acetyltransferases CBP and GCN5. A, histone deacetylation upon treatment of MDA-MB-231 cells with 200 nmol/L of TSA reactivated KLK6 transcription by recruiting proteins from the basal transcription machinery on KLK6 chromatin. Crosslinked chromatin prepared from MDA-MB-468 and MDA-MB-231 cells was immunoprecipitated with specific antibodies (right). Increasing crosslinking times (top). Immunoprecipitates were subjected to PCR using a primer pair spanning the region containing the methylated CpGs (Fig. 1D). Aliquots of chromatin taken before immunoprecipitation were used as “Input” controls. B, recruitment of MeCP2, histone acetyltransferases, and acetylation status of H4 at the KLK6 CpG-containing chromatin was shown by ChIP. In vivo binding of Ac-H4, MeCP2, CBP, and GCN5 was examined in KLK6-positive MDA-MB-468 and in KLK6-negative MDA-MB-231 cells (left). KLK6 chromatin is enriched in Ac-H4 upon HDAC inhibition with TSA (right). C, release of the MeCP2 repressor and marked increase of Ac-H4 binding to KLK6 chromatin following treatment of T47D cells with 100 μmol/L of 5-aza-dC.

Close modal

Methylation-dependent binding of MeCP2 mediates the repression of KLK6 expression in breast cancer cells. To investigate the recruitment events associated with constitutive and inducible KLK6 transcription in breast cancer cells, we performed a series of ChIP experiments using chromatin prepared from MDA-MB-468 and MDA-MB-231 cells and antibodies against the acetylated tail of histone H4, MeCP2, and histone acetyltransferases, CBP and GCN5. Figure 2B and C shows that constitutive and inducible expression of KLK6 positively correlates with histone H4 tail acetylation. KLK6 upstream sequences containing the identified methylated CpG dinucleotides are enriched in MDA-MB-468 chromatin that was immunoprecipitated with the α-acetyl histone H4 compared with MDA-MB-231 chromatin (Fig. 2B). Additionally, both TSA and 5-aza-dC treatment resulted in increased histone H4 tail acetylation in the KLK6 promoter in MDA-MB-231 (Fig. 2B) and in T47D cells (Fig. 2C), respectively. This enhanced histone H4 acetylation was accompanied by the presence of the CBP coactivator and TATA-binding protein basal transcription factor to KLK6 promoter in MDA-MB-468 and TSA-treated MDA-MB-231 cells. CBP is a histone acetyltransferase that, along with GCN5, could be responsible for the enhanced histone H4 acetylation that we observed when the gene is expressed. Finally, we observed an inverse correlation between binding of the methyl-binding protein MeCP2 to KLK6 upstream sequences and expression of KLK6. MeCP2 binding is not detected in KLK6 upstream sequences in MDA-MB-468–expressing cells whereas MeCP2 departs in TSA-treated MDA-MB-231 and 5-aza-dC–treated T47D cells, respectively. Taken together, these data show that TSA and 5-aza-dC treatment of cancer cells induce KLK6 gene transcription and that this induction correlates with departure of MeCP2 repressor from the KLK6 promoter.

Selection of KLK6-transfected cells. The epigenetic silencing of KLK6 in metastatic tumor cells that was described here points to a putative tumor suppressor function of the encoded protein. Although loss of function of tumor suppressor genes through genetic mutations is a hallmark in human cancers, hypermethylation of genomic DNA that usually, but not exclusively, occurs in promoter sequences provides a frequent mechanism of transcriptional silencing and functional disruption of known tumor suppressor genes in cancer cells (13, 1722). To test whether inactivation of KLK6 expression by genomic DNA hypermethylation, and thereafter, lack of KLK6 function provides a growth advantage to metastatic breast tumor cells, the MDA-MB-231 cell line was stably transfected with an expression construct that directs the synthesis of preproKLK6. Stably transfected cells were selected based on the concentrations of secreted KLK6 protein determined by ELISA (Table 1) and established in cultures.

Table 1.

In vitro growth characteristics of MDA-MB-231 parental, mock-transfected, and KLK6-transfected clones stably expressing varying concentrations of KLK6

Cell lineKLK6 (μg/L)PDT (h)Saturation density (106/cm2)
Parental ND 33.63 ± 1.00 1.05 ± 0.14 
Mock ND 28.90 ± 0.60 1.06 ± 0.17 
C5wt >100 33.90 ± 0.80 1.02 ± 0.20 
C6wt 2.5 41.35 ± 2.47 0.64 ± 0.10 
C11wt 1.0 43.63 ± 1.11 0.55 ± 0.05 
C12wt 28 40.13 ± 0.98 0.75 ± 0.15 
Cell lineKLK6 (μg/L)PDT (h)Saturation density (106/cm2)
Parental ND 33.63 ± 1.00 1.05 ± 0.14 
Mock ND 28.90 ± 0.60 1.06 ± 0.17 
C5wt >100 33.90 ± 0.80 1.02 ± 0.20 
C6wt 2.5 41.35 ± 2.47 0.64 ± 0.10 
C11wt 1.0 43.63 ± 1.11 0.55 ± 0.05 
C12wt 28 40.13 ± 0.98 0.75 ± 0.15 

Abbreviations: ND, not detected; PDT, population doubling time.

Stable expression of KLK6 reduces proliferation rates, motility, and anchorage-independent growth of MDA-MB-231 cells. The growth rates of parental and mock-transfected cells were compared with that of KLK6-transfected cells. As shown in Fig. 3A and Table 1, expression of KLK6 protein at physiologic levels (clones: C6wt, C11wt, and C12wt) resulted in reduced proliferation rates of MDA-MB-231 cells. Interestingly, KLK6-transfected cells grew to lower saturation densities than parental or mock controls (Table 1). Based on previous observations that a small subset of primary breast tumors highly overexpress (e.g., 50-fold to 100-fold higher than normal) KLK6 mRNA and protein (1), a KLK6-overexpressing clone (C5wt) was selected that was found to proliferate with exactly the same rate and reached similar saturation density as parental cells (Table 1). The motility of parental, mock-transfected, and KLK6-transfected MDA-MB-231 cells (Fig. 3B) was assessed with the wound scratch (healing) assay (23). Parental and mock-transfected cells were capable of closing the wound almost completely within 24 hours, whereas cells of KLK6-expressing MDA-MB-231 clones were less motile and showed significantly slower rates of wound-healing (Fig. 3B). Finally, the effect of KLK6 expression on the tumorigenicity of MDA-MB-231 was assessed in vitro by soft agar assay (12). As shown in Fig. 3C, parental and mock controls yielded numerous colonies in a period of 3 weeks, whereas a very small number of smaller-sized colonies were obtained for KLK6-expressing clones C6wt, C12wt, and C11wt (data not shown), indicating that KLK6 protein acts to inhibit the anchorage-independent growth of tumor cells. The growth of C5wt cells in soft agar, although significant, was suppressed to a lesser extent (Fig. 3C). Taken together, our data shows that KLK6 reverses the malignant phenotype of MDA-MB-231 cells and supports the idea that KLK6 might play a tumor suppressor role in breast cancer.

Figure 3.

Re-expression of KLK6 by stable cDNA transfection results in the suppression of the malignant phenotype of MDA-MB-231 breast tumor cells. A, KLK6 reduces the proliferation rates and saturation density of MDA-MB-231 cells. B, KLK6 reduces the motility of MDA-MB-231 cells as assessed by the wound-healing assay. A representative experiment of at least three replicates. C, transfection of KLK6 cDNA in MDA-MB-231 remarkably suppressed anchorage-independent growth of MDA-MB-231 cells. Inhibition of anchorage-independent growth was not as prominent for C5wt cells that highly overexpress KLK6. Results and corresponding SDs were derived from three independent experiments.

Figure 3.

Re-expression of KLK6 by stable cDNA transfection results in the suppression of the malignant phenotype of MDA-MB-231 breast tumor cells. A, KLK6 reduces the proliferation rates and saturation density of MDA-MB-231 cells. B, KLK6 reduces the motility of MDA-MB-231 cells as assessed by the wound-healing assay. A representative experiment of at least three replicates. C, transfection of KLK6 cDNA in MDA-MB-231 remarkably suppressed anchorage-independent growth of MDA-MB-231 cells. Inhibition of anchorage-independent growth was not as prominent for C5wt cells that highly overexpress KLK6. Results and corresponding SDs were derived from three independent experiments.

Close modal

Identification of molecular alterations caused by KLK6. Differential proteomic profiling of parental, mock-transfected, and KLK6-transfected MDA-MB-231 cells was applied to the identification of potential substrates for KLK6. Figure 4A shows a representative pattern of two-dimensional electrophoresis of cell lysates. Proteins displaying differential expression between parental (data not shown), mock-transfected, and KLK6-transfected C12wt cells are shown (Fig. 4A; Supplementary Table S1). Unexpectedly, we found that vimentin, a mesenchymal marker, was dramatically down-regulated in KLK6 transfectants as compared with mock and parental controls (mean, 9-fold reduction). Western blot analysis confirmed that vimentin levels are significantly decreased in KLK6 transfectants (Fig. 4B). Because RT-PCR showed no change in vimentin mRNA, reduction of vimentin concentration must occur posttranslationally. Down-regulation of vimentin upon expression of KLK6 suggests a possible “mesenchymal-to-epithelial–like” (MET-like) transition that would be compatible with a tumor suppressor role (24). Interestingly, significant up-regulation of the epithelial markers cytokeratin 8 (3-fold) and cytokeratin 19 (10-fold), as well as of calreticulin (3-fold), was detected in KLK6-transfected cells. However, expression of E-cadherin, which is an established epithelial marker (24), was not induced in KLK6 transfectants, as shown by RT-PCR and Western blot (Fig. 4B).

Figure 4.

Molecular changes in the intracellular proteome accompanied by reconstitution of KLK6 expression. A, lysates from mock and C12wt cells were analyzed by two-dimensional PAGE. Circles, differentially expressed proteins (also shown in Supplementary Table S1). B, down-regulation of vimentin was confirmed by Western blot analysis, whereas no change in mRNA levels was detected by RT-PCR. Protein and mRNA levels of E-cadherin were not altered.

Figure 4.

Molecular changes in the intracellular proteome accompanied by reconstitution of KLK6 expression. A, lysates from mock and C12wt cells were analyzed by two-dimensional PAGE. Circles, differentially expressed proteins (also shown in Supplementary Table S1). B, down-regulation of vimentin was confirmed by Western blot analysis, whereas no change in mRNA levels was detected by RT-PCR. Protein and mRNA levels of E-cadherin were not altered.

Close modal

KLK6 inhibits tumor formation in SCID mice. Parental MDA-MB-231 and mock-transfected cells grew to palpable tumors 6 to 7 weeks after implantation into the mammary fat-pads of SCID mice (Table 2; Supplementary Fig. S2, top). Tumors developed rapidly and exceeded a volume of 1,000 mm3 after 4 months. In one case, the tumor exceeded a volume of 6,000 mm3. In contrast, KLK6-transfected cells showed significant inhibition of tumor growth in vivo. More specifically, clones C11wt and C12wt did not develop tumors even after 4 months following implantation, whereas clone C6wt grew tumors, although with a significantly delayed onset and a slower proliferation rate when compared with parental or mock controls (Table 2; Supplementary Fig. S2, bottom). The tumor volume was significantly smaller for C6wt cells (mean, 20 mm3; day 77) when compared with parental and mock controls (mean, 200 mm3; day 77; Table 2). Taken together, these results indicate that re-expression of KLK6 at physiologic concentrations inhibits tumor formation in vivo and support a tumor suppressor role for this protein. Interestingly, when KLK6 is produced at abnormally high levels, e.g., >100 and 400 μg/L, as in C5wt and MDA-MB-468 tumor cells, respectively, no inhibition of tumor growth was observed. We found that C5wt cells form tumors with the same efficiency as parental and mock controls in terms of onset, growth rates, frequency, and tumor volume at the end point (Table 2; Supplementary Fig. S2). Notably, in two C5wt-injected sites, the final tumor volume exceeded 6,000 mm3, as also observed for parental MDA-MB-231 cells in one case. Similarly, the onset of MDA-MB-468 tumor growth in SCID mice was observed in only 3 weeks and tumors continued to grow very efficiently (in six of six sites injected; data not shown).

Table 2.

Tumor formation in SCID mice

Cell lineOnset of tumor formation (wk)No. of sites injectedSites which developed tumors at 77 d (%)Size of primary tumors at 77 d (mm3)Sites which developed tumors at 96 d (%)Size of primary tumors at 96 d (mm3)
Parental 15 100 227 ± 68 100 1,357 ± 366 
Mock 6–7 16 100 270 ± 100 100 Sacrificed 
C5wt 6–7 18 100 292 ± 274 100 1,429 ± 995 
C6wt 7–8 18 72 21 ± 13 89 430 ± 122 
C11wt No tumor 10 
C12wt No tumor 24 
Cell lineOnset of tumor formation (wk)No. of sites injectedSites which developed tumors at 77 d (%)Size of primary tumors at 77 d (mm3)Sites which developed tumors at 96 d (%)Size of primary tumors at 96 d (mm3)
Parental 15 100 227 ± 68 100 1,357 ± 366 
Mock 6–7 16 100 270 ± 100 100 Sacrificed 
C5wt 6–7 18 100 292 ± 274 100 1,429 ± 995 
C6wt 7–8 18 72 21 ± 13 89 430 ± 122 
C11wt No tumor 10 
C12wt No tumor 24 

Kallikrein-related peptidases participate in proteolytic cascades that regulate important physiologic and pathophysiologic processes (25). Aberrant expression of KLK6 has been associated with human breast and ovarian cancers, although the molecular mechanisms underlying this dysregulation have not been described. Here, we show that 5-aza-dC induces KLK6 expression in KLK6-negative breast cancer cell lines. TSA reactivates KLK6 only in MDA-MB-231 cells. Although KLK6 lacks a typical CpG island, the methylation status of distinct CpG dinucleotides located in KLK6 proximal promoter correlated with KLK6 expression. We found that tumor-specific loss of KLK6 expression results from hypermethylation of these CpGs, whereas their complete demethylation is associated with the constitutive KLK6 expression observed in a subset of breast tumors.

Recruitment events to KLK6 proximal promoter were studied by ChIP in MDA-MB-468 cells, in which KLK6 expression is constitutive, and during induction of KLK6 expression in TSA-treated MDA-MB-231 and in 5-aza-dC–treated T47D cells. It was shown that the mechanisms of constitutive and inducible transcription of KLK6 are likely different and that CpG methylation–mediated KLK6 silencing in cancer cells is associated with the formation of transcriptional repression complexes, e.g., recruitment of MeCP2 and localized deacetylation of histone H4. The identification of the MeCP2 repressor in KLK6 chromatin is compatible with the lack of CpG islands because MeCP2-mediated repression does not require a CpG island. Interestingly, MeCP2 has been shown to repress transcription by binding even to a single methylCpG (26). Methylated DNA and MeCP2 recruit corepressor complexes that mediate repression through deacetylation of core histones (27), with consequent compaction of DNA into the heterochromatin. MeCP2 can displace histone H1 from preassembled chromatin that contains methylCpG and recruits HDACs to suppress gene expression (2730). Our data shows that 5-aza-dC treatment of breast cancer cells induces KLK6 transcription and that this induction correlates with the departure of the MeCP2 transcriptional repressor from the KLK6 promoter. Surprisingly, activation of KLK6 by TSA also forces MeCP2 to vacate the promoter. It could be either that MeCP2 protein binding is incompatible with histone acetylation on this promoter or that TSA may cause site-specific DNA demethylation (31). In conclusion, methylation-dependent binding of MeCP2 and the formation of repressive chromatin inhibit the transcription of KLK6 by interfering with the recruitment and function of the basal transcription machinery.

Epigenetic silencing of KLK6 in breast cancer cells suggested that the encoded protein could normally play a tumor suppressor role. Therefore, we stably transfected the KLK6 cDNA into nonexpressing MDA-MB-231 breast tumor cell line and investigated potential effects on the tumor cell phenotype. We found that re-expression of KLK6 at physiologic concentrations, as those found in breast cyst fluid from normal humans, e.g., 1 to 30 μg/L (mean, 10 μg/L; ref. 32), reversed the tumorigenicity of MDA-MB-231 cells in vitro. Most importantly, re-expression of KLK6 resulted in the repression of in vivo orthotopic tumor formation when MDA-MB-231 cells were implanted in SCID mice, indicating a tumor suppressor role for KLK6 in breast cancer. Notably, in C5wt and MDA-MB-468 tumor cells, in which KLK6 is produced at abnormally high levels, no inhibition of tumor growth was observed. Here, we showed that this constitutive overexpression of KLK6 in MDA-MB-468 and 21PT (data not shown) primary breast tumor cells is associated with complete demethylation of the identified CpGs. Overexpression of KLK6 in cancer cells might also result from gene amplification, as reported for a subset of ovarian tumors (33). Possibly, such abnormal tumor-associated overproduction of KLK6 may stimulate tumor growth by activating protease-activated receptor 2 signaling (34), given the fact that both MDA-MB-468 and MDA-MB-231 cells express protease-activated receptor 2 (35). Indeed, it was shown recently that KLK6 can activate protease-activated receptor 2 at very high concentrations (1.6 nmol/L; ref. 34) that are comparable to those produced by C5wt (2.5 nmol/L) and MDA-MB-468 (10 nmol/L). In contrast, in normal HMECs and in KLK6 transfectants (C6wt, C11wt, and C12wt), the concentration of KLK6 lies in the range 25 to 600 pmol/L. Validation of protease-activated receptor 2 activation in KLK6-overexpressing tumors will be required to confirm this hypothesis. Transient overexpression of KLK6, which was originally observed in 21PT primary breast tumor as compared with its corresponding metastasis (21MT-1; ref. 1), might support tumor growth at the primary site. On the other hand, reactivation of KLK6 expression at normal levels inhibits tumor progression, as shown here, by inducing MET-like tumor reversion.

Indeed, differential proteomic profiling of parental, mock-transfected, and KLK6-transfected MDA-MB-231 cells revealed that restoration of KLK6 results in extensive down-regulation of the major mesenchymal marker vimentin that is characteristic of aggressive forms of breast cancer (36), and concomitant up-regulation of epithelial markers such as cytokeratin 8 and cytokeratin 19, and of calreticulin, a protein with a molecular chaperone function and antiangiogenic activity (37). Down-regulation of vimentin is associated with reduced tumor aggressiveness (38). In addition, up-regulation of cytokeratin 8, which is usually not expressed in metastatic breast cancer cells (39), and reduced motility observed in KLK6-transfected cells are both compatible with MET-like tumor reversion (24). Therefore, KLK6 likely promotes a MET-like transition, a function that merits further thorough investigation. Vimentin was also markedly repressed in KLK6-overexpressing C5wt and MDA-MB-468 cells. It should be noted, however, that we could not detect a significant up-regulation of E-cadherin that represents the major epithelial marker or “caretaker” of the epithelial phenotype (24). How KLK6 expression results in down-regulation of vimentin is unknown. Direct cleavage of vimentin by KLK6 is unlikely because KLK6 is an extracellular protease. Construction of the KLK6 interactome revealed that KLK6 might act via the transforming growth factor-β1 (TGF-β1) pathway to reduce TGF-β1 levels by degrading the extracellular protein fibronectin (40). Fibronectin increases steady-state mRNA levels of TGF-β1 that is known to up-regulate vimentin and to down-regulate cytokeratin 8 and calreticulin. Indeed, we have shown that TGF-β1 levels reverse-correlate with KLK6 expression in TGF-β1–treated breast tumor cells.5

5

Our unpublished data.

Proteases were considered to act as tumor promoters by degrading components of the extracellular matrix. Accumulating evidence indicates that specific proteases inhibit early tumor growth or/and progression (41). KLK6 has been implicated in the induction of differentiation of colon (42) and squamous cell carcinomas (43), and in the inhibition of angiogenesis (6), compatible with its function as a tumor suppressor.

In summary, it was shown for the first time that KLK6 normally exerts a tumor-protective function against breast cancer that is likely mediated by inhibition of epithelial-to-mesenchymal transition, and that epigenetic mechanisms underlie its loss-of-function in breast tumor cells. Pharmacologic modulation of KLK6 gene expression described here may be exploited therapeutically.

No potential conflicts of interest were disclosed.

Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/).

Grant support: K. Karatheodoris (C.186) provided by the Research Committee of the University of Patras and PENED2003 (03EΔ430) cofunded by the E.U. European Social Fund (75%) and the Greek Ministry of Development-GSRT (25%).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We thank Eleftherios P. Diamandis and Antoninus Soosaipillai for ELISA determinations, Nikos Youroukos for his help with animal experiments, and Evi Lianidou and Vassilis Georgoulias for providing the clinical samples.

1
Anisowicz A, Sotiropoulou G, Stenman G, Mok SC, Sager R. A novel protease homolog differentially expressed in breast and ovarian cancer.
Mol Med
1996
;
2
:
624
–6.
2
Borgoño CA, Diamandis EP. The emerging roles of human tissue kallikreins in cancer.
Nat Rev Cancer
2004
;
4
:
876
–90.
3
Bernett MJ, Blaber SI, Scarisbrick IA, Dhanarajan P, Thompson SM, Blaber M. Crystal structure and biochemical characterization of human kallikrein 6 reveals that a trypsin-like kallikrein is expressed in the central nervous system.
J Biol Chem
2002
;
277
:
24562
–70.
4
Scarisbrick IA, Blaber SI, Lucchinetti CF, Genain CP, Blaber M, Rodriguez M. Activity of a newly identified serine protease in CNS demyelination.
Brain
2002
;
125
:
1283
–96.
5
Blaber SI, Ciric B, Christophi GP, et al. Targeting kallikrein 6 proteolysis attenuates CNS inflammatory disease.
FASEB J
2004
;
18
:
920
–2.
6
Bayés A, Tsetsenis T, Ventura S, Vendrell J, Avilés FX, Sotiropoulou G. Human kallikrein 6 activity is regulated via an autoproteolytic mechanism of activation inactivation.
Biol Chem
2004
;
385
:
517
–24.
7
Gomis-Ruth FX, Bayés A, Sotiropoulou G, et al. The structure of human prokallikrein 6 reveals a novel activation mechanism for the kallikrein family.
J Biol Chem
2002
;
277
:
27273
–81.
8
Pampalakis G, Kurlender L, Diamandis EP, Sotiropoulou G. Cloning and characterization of novel isoforms of the human kallikrein 6 gene.
Biochem Biophys Res Commun
2004
;
320
:
54
–61.
9
Christophi GP, Isackson PJ, Blaber S, Blaber M, Rodriguez M, Scarisbrick IA. Distinct promoters regulate tissue-specific and differential expression of kallikrein 6 in CNS demyelinating disease.
J Neurochem
2004
;
91
:
1439
–49.
10
Agalioti T, Chen G, Thanos D. Deciphering the transcriptional histone acetylation code for a human gene.
Cell
2002
;
111
:
381
–92.
11
Diamandis EP, Yousef GM, Soosaipillai AR, et al. Immunofluorometric assay of human kallikrein 6 (zyme/protease M/neurosin) and preliminary clinical applications.
Clin Biochem
2000
;
33
:
369
–75.
12
Goyal J, Smith KM, Cowan JM, Wazer DE, Lee SW, Band V. The role for NES1 serine protease as a novel tumor suppressor.
Cancer Res
1998
;
58
:
4782
–6.
13
Li B, Goyal J, Dhar S, et al. CpG methylation as a basis for breast tumor-specific loss of NES1/kallikrein 10 expression.
Cancer Res
2001
;
61
:
8014
–21.
14
Jones PA, Baylin SB. The fundamental role of epigenetic events in cancer.
Nat Rev Genet
2002
;
3
:
415
–28.
15
Pampalakis G, Sotiropoulou G. Multiple mechanisms underlie the aberrant expression of the human kallikrein 6 gene in breast cancer.
Biol Chem
2006
;
387
:
773
–82.
16
Futscher BW, Oshiro MM, Wozniak RJ, et al. Role for DNA methylation in the control of cell type specific maspin expression.
Nat Genet
2002
;
31
:
175
–9.
17
Herman JG, Latif F, Weng Y, et al. Silencing of the VHL tumor-suppressor gene by DNA methylation in renal carcinoma.
Proc Natl Acad Sci U S A
1994
;
91
:
9700
–4.
18
Merlo A, Herman JG, Mao L, et al. 5′ CpG island methylation is associated with transcriptional silencing of the tumour suppressor p16/CDKN2/MTS1 in human cancers.
Nat Med
1995
;
1
:
686
–92.
19
Gonzalez-Zulueta M, Bender CM, Yang AS, et al. Methylation of the 5′ CpG island of the p16/CDKN2 tumor suppressor gene in normal and transformed human tissues correlates with gene silencing.
Cancer Res
1995
;
55
:
4531
–5.
20
Herman JG, Umar A, Polyak K, et al. Incidence and functional consequences of hMLH1 promoter hypermethylation in colorectal carcinoma.
Proc Natl Acad Sci U S A
1998
;
95
:
6870
–5.
21
Esteller M, Silva JM, Dominguez G, et al. Promoter hypermethylation and BRCA1 inactivation in sporadic breast and ovarian tumors.
J Natl Cancer Inst
2000
;
92
:
564
–9.
22
Nass SJ, Herman JG, Gabrielson E, et al. Aberrant methylation of the estrogen receptor and E-cadherin 5′ CpG islands increases with malignant progression in human breast cancer.
Cancer Res
2000
;
60
:
4346
–8.
23
Rodriguez LG, Wu X, Guan JL. Wound-healing assay.
Methods Mol Biol
2005
;
294
:
23
–9.
24
Thiery JP. Epithelial-mesenchymal transitions in tumour progression.
Nat Rev Cancer
2002
;
2
:
442
–54.
25
Pampalakis G, Sotiropoulou G. Tissue kallikrein proteolytic cascade pathways in normal physiology and cancer.
Biochim Biophys Acta
2007
;
1776
:
22
–31.
26
Nan X, Meehan RR, Bird A. Dissection of the methyl-CpG binding domain from the chromosomal protein MeCP2.
Nucleic Acids Res
1993
;
21
:
4886
–92.
27
Jones PL, Veenstra GJ, Wade PA, et al. Methylated DNA and MeCP2 recruit histone deacetylase to repress transcription.
Nat Genet
1998
;
19
:
187
–91.
28
Eden S, Hashimshony T, Keshet I, Cedar H, Thorne AW. DNA methylation models histone acetylation.
Nature
1998
;
394
:
842
.
29
Nan X, Ng HH, Johnson CA, et al. Transcriptional repression by the methyl-CpG-binding protein MeCP2 involves a histone deacetylase complex.
Nature
1998
;
393
:
386
–9.
30
Fuks F, Hurd PJ, Wolf D, Nan X, Bird AP, Kouzarides T. The methyl-CpG-binding protein MeCP2 links DNA methylation to histone methylation.
J Biol Chem
2003
;
278
:
4035
–40.
31
Gan Y, Shen YH, Wang J, et al. Role of histone deacetylation in cell-specific expression of endothelial nitric-oxide synthase.
J Biol Chem
2005
;
280
:
16467
–75.
32
Shaw JL, Diamandis EP. Distribution of 15 human kallikreins in tissues and biological fluids.
Clin Chem
2007
;
53
:
1423
–32.
33
Ni X, Zhang W, Huang KC, et al. Characterization of human kallikrein 6/protease M expression in ovarian cancer.
Br J Cancer
2004
;
91
:
725
–31.
34
Oikonomopoulou K, Hansen KK, Saifeddine M, et al. Proteinase-activated receptors, targets for kallikrein signaling.
J Biol Chem
2006
;
281
:
32095
–112.
35
Ge L, Shenoy SK, Lefkowitz RJ, DeFea K. Constitutive protease-activated receptor-2-mediated migration of MDA-MB-231 breast cancer cells requires both β-arrestin-1 and -2.
J Biol Chem
2004
;
53
:
55419
–24.
36
Yang J, Mani SA, Donaher JL, et al. Twist, a master regulator of morphogenesis, plays an essential role in tumor metastasis.
Cell
2004
;
117
:
927
–39.
37
Cheng WF, Hung CF, Chen CA, et al. Characterization of DNA vaccines encoding the domains of calreticulin for their ability to elicit tumor-specific immunity and antiangiogenesis.
Vaccine
2005
;
23
:
3864
–74.
38
Buhler H, Schaller G. Transfection of keratin 18 gene in human breast cancer cells causes induction of adhesion proteins and dramatic regression of malignancy in vitro and in vivo.
Mol Cancer Res
2005
;
3
:
365
–71.
39
Willipinski-Stapelfeldt B, Riethdorf S, Assmann V, et al. Changes in cytoskeletal protein composition indicative of an epithelial-mesenchymal transition in human micrometastatic and primary breast carcinoma cells.
Clin Cancer Res
2005
;
11
:
8006
–14.
40
Pampalakis G, Arampatzidou M, Amoutzias G, Kossida S, Sotiropoulou G. Identification and analysis of mammalian KLK6 orthologue genes for prediction of physiological substrates.
Comput Biol Chem
2008
;
32
:
111
–21.
41
Lopéz-Otín C, Matrisian LM. Emerging roles of proteases in tumour suppression.
Nat Rev Cancer
2007
;
7
:
800
–8.
42
Palmer HG, Sanchez-Carbayo M, Ordonez-Moran P, Larriba MJ, Cordon-Cardo C, Munoz A. Genetic signatures of differentiation induced by 1α,25-dihydroxy-vitamin D3 in human colon cancer cells.
Cancer Res
2003
;
63
:
7799
–806.
43
Lin R, Nagai Y, Sladek R, et al. Expression profiling in squamous carcinoma cells reveals pleiotropic effects of vitamin D3 analog EB1089 signaling on cell proliferation, differentiation, and immune system regulation.
Mol Endocrinol
2003
;
16
:
1243
–56.